Add custom Timestamp type for sample times.
So far we've been using Go's native time.Time for anything related to sample
timestamps. Since the range of time.Time is much bigger than what we need, this
has created two problems:
- there could be time.Time values which were out of the range/precision of the
time type that we persist to disk, therefore causing incorrectly ordered keys.
One bug caused by this was:
https://github.com/prometheus/prometheus/issues/367
It would be good to use a timestamp type that's more closely aligned with
what the underlying storage supports.
- sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit
Unix timestamp (possibly even a 32-bit one). Since we store samples in large
numbers, this seriously affects memory usage. Furthermore, copying/working
with the data will be faster if it's smaller.
*MEMORY USAGE RESULTS*
Initial memory usage comparisons for a running Prometheus with 1 timeseries and
100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my
tests, this advantage for some reason decreased a bit the more samples the
timeseries had (to 5-7% for millions of samples). This I can't fully explain,
but perhaps garbage collection issues were involved.
*WHEN TO USE THE NEW TIMESTAMP TYPE*
The new clientmodel.Timestamp type should be used whenever time
calculations are either directly or indirectly related to sample
timestamps.
For example:
- the timestamp of a sample itself
- all kinds of watermarks
- anything that may become or is compared to a sample timestamp (like the timestamp
passed into Target.Scrape()).
When to still use time.Time:
- for measuring durations/times not related to sample timestamps, like duration
telemetry exporting, timers that indicate how frequently to execute some
action, etc.
Change-Id: I253a467388774280c10400fda122369ff77c1730
2013-10-29 19:36:58 +04:00
|
|
|
// Copyright 2013 Prometheus Team
|
|
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
// you may not use this file except in compliance with the License.
|
|
|
|
// You may obtain a copy of the License at
|
|
|
|
//
|
|
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
//
|
|
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
// See the License for the specific language governing permissions and
|
|
|
|
// limitations under the License.
|
|
|
|
|
|
|
|
package model
|
|
|
|
|
|
|
|
import (
|
|
|
|
"fmt"
|
|
|
|
native_time "time"
|
|
|
|
)
|
|
|
|
|
|
|
|
// TODO(julius): Should this use milliseconds/nanoseconds instead? This is
|
|
|
|
// mostly hidden from the user of these types when using the
|
|
|
|
// methods below, so it will be easy to change this later
|
|
|
|
// without requiring significant user code changes.
|
|
|
|
|
2014-04-25 14:50:18 +04:00
|
|
|
// Timestamp is the number of seconds since the epoch (1970-01-01 00:00 UTC)
|
|
|
|
// without leap seconds.
|
Add custom Timestamp type for sample times.
So far we've been using Go's native time.Time for anything related to sample
timestamps. Since the range of time.Time is much bigger than what we need, this
has created two problems:
- there could be time.Time values which were out of the range/precision of the
time type that we persist to disk, therefore causing incorrectly ordered keys.
One bug caused by this was:
https://github.com/prometheus/prometheus/issues/367
It would be good to use a timestamp type that's more closely aligned with
what the underlying storage supports.
- sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit
Unix timestamp (possibly even a 32-bit one). Since we store samples in large
numbers, this seriously affects memory usage. Furthermore, copying/working
with the data will be faster if it's smaller.
*MEMORY USAGE RESULTS*
Initial memory usage comparisons for a running Prometheus with 1 timeseries and
100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my
tests, this advantage for some reason decreased a bit the more samples the
timeseries had (to 5-7% for millions of samples). This I can't fully explain,
but perhaps garbage collection issues were involved.
*WHEN TO USE THE NEW TIMESTAMP TYPE*
The new clientmodel.Timestamp type should be used whenever time
calculations are either directly or indirectly related to sample
timestamps.
For example:
- the timestamp of a sample itself
- all kinds of watermarks
- anything that may become or is compared to a sample timestamp (like the timestamp
passed into Target.Scrape()).
When to still use time.Time:
- for measuring durations/times not related to sample timestamps, like duration
telemetry exporting, timers that indicate how frequently to execute some
action, etc.
Change-Id: I253a467388774280c10400fda122369ff77c1730
2013-10-29 19:36:58 +04:00
|
|
|
type Timestamp int64
|
|
|
|
|
|
|
|
const (
|
2014-04-25 14:50:18 +04:00
|
|
|
// MinimumTick is the minimum supported time resolution. This has to be
|
|
|
|
// at least native_time.Second in order for the code below to work.
|
Add custom Timestamp type for sample times.
So far we've been using Go's native time.Time for anything related to sample
timestamps. Since the range of time.Time is much bigger than what we need, this
has created two problems:
- there could be time.Time values which were out of the range/precision of the
time type that we persist to disk, therefore causing incorrectly ordered keys.
One bug caused by this was:
https://github.com/prometheus/prometheus/issues/367
It would be good to use a timestamp type that's more closely aligned with
what the underlying storage supports.
- sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit
Unix timestamp (possibly even a 32-bit one). Since we store samples in large
numbers, this seriously affects memory usage. Furthermore, copying/working
with the data will be faster if it's smaller.
*MEMORY USAGE RESULTS*
Initial memory usage comparisons for a running Prometheus with 1 timeseries and
100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my
tests, this advantage for some reason decreased a bit the more samples the
timeseries had (to 5-7% for millions of samples). This I can't fully explain,
but perhaps garbage collection issues were involved.
*WHEN TO USE THE NEW TIMESTAMP TYPE*
The new clientmodel.Timestamp type should be used whenever time
calculations are either directly or indirectly related to sample
timestamps.
For example:
- the timestamp of a sample itself
- all kinds of watermarks
- anything that may become or is compared to a sample timestamp (like the timestamp
passed into Target.Scrape()).
When to still use time.Time:
- for measuring durations/times not related to sample timestamps, like duration
telemetry exporting, timers that indicate how frequently to execute some
action, etc.
Change-Id: I253a467388774280c10400fda122369ff77c1730
2013-10-29 19:36:58 +04:00
|
|
|
MinimumTick = native_time.Second
|
2014-04-25 14:50:18 +04:00
|
|
|
// second is the timestamp duration equivalent to one second.
|
Add custom Timestamp type for sample times.
So far we've been using Go's native time.Time for anything related to sample
timestamps. Since the range of time.Time is much bigger than what we need, this
has created two problems:
- there could be time.Time values which were out of the range/precision of the
time type that we persist to disk, therefore causing incorrectly ordered keys.
One bug caused by this was:
https://github.com/prometheus/prometheus/issues/367
It would be good to use a timestamp type that's more closely aligned with
what the underlying storage supports.
- sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit
Unix timestamp (possibly even a 32-bit one). Since we store samples in large
numbers, this seriously affects memory usage. Furthermore, copying/working
with the data will be faster if it's smaller.
*MEMORY USAGE RESULTS*
Initial memory usage comparisons for a running Prometheus with 1 timeseries and
100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my
tests, this advantage for some reason decreased a bit the more samples the
timeseries had (to 5-7% for millions of samples). This I can't fully explain,
but perhaps garbage collection issues were involved.
*WHEN TO USE THE NEW TIMESTAMP TYPE*
The new clientmodel.Timestamp type should be used whenever time
calculations are either directly or indirectly related to sample
timestamps.
For example:
- the timestamp of a sample itself
- all kinds of watermarks
- anything that may become or is compared to a sample timestamp (like the timestamp
passed into Target.Scrape()).
When to still use time.Time:
- for measuring durations/times not related to sample timestamps, like duration
telemetry exporting, timers that indicate how frequently to execute some
action, etc.
Change-Id: I253a467388774280c10400fda122369ff77c1730
2013-10-29 19:36:58 +04:00
|
|
|
second = int64(native_time.Second / MinimumTick)
|
|
|
|
)
|
|
|
|
|
|
|
|
// Equal reports whether two timestamps represent the same instant.
|
|
|
|
func (t Timestamp) Equal(o Timestamp) bool {
|
|
|
|
return t == o
|
|
|
|
}
|
|
|
|
|
|
|
|
// Before reports whether the timestamp t is before o.
|
|
|
|
func (t Timestamp) Before(o Timestamp) bool {
|
|
|
|
return t < o
|
|
|
|
}
|
|
|
|
|
2014-04-25 14:50:18 +04:00
|
|
|
// After reports whether the timestamp t is after o.
|
Add custom Timestamp type for sample times.
So far we've been using Go's native time.Time for anything related to sample
timestamps. Since the range of time.Time is much bigger than what we need, this
has created two problems:
- there could be time.Time values which were out of the range/precision of the
time type that we persist to disk, therefore causing incorrectly ordered keys.
One bug caused by this was:
https://github.com/prometheus/prometheus/issues/367
It would be good to use a timestamp type that's more closely aligned with
what the underlying storage supports.
- sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit
Unix timestamp (possibly even a 32-bit one). Since we store samples in large
numbers, this seriously affects memory usage. Furthermore, copying/working
with the data will be faster if it's smaller.
*MEMORY USAGE RESULTS*
Initial memory usage comparisons for a running Prometheus with 1 timeseries and
100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my
tests, this advantage for some reason decreased a bit the more samples the
timeseries had (to 5-7% for millions of samples). This I can't fully explain,
but perhaps garbage collection issues were involved.
*WHEN TO USE THE NEW TIMESTAMP TYPE*
The new clientmodel.Timestamp type should be used whenever time
calculations are either directly or indirectly related to sample
timestamps.
For example:
- the timestamp of a sample itself
- all kinds of watermarks
- anything that may become or is compared to a sample timestamp (like the timestamp
passed into Target.Scrape()).
When to still use time.Time:
- for measuring durations/times not related to sample timestamps, like duration
telemetry exporting, timers that indicate how frequently to execute some
action, etc.
Change-Id: I253a467388774280c10400fda122369ff77c1730
2013-10-29 19:36:58 +04:00
|
|
|
func (t Timestamp) After(o Timestamp) bool {
|
|
|
|
return t > o
|
|
|
|
}
|
|
|
|
|
|
|
|
// Add returns the Timestamp t + d.
|
|
|
|
func (t Timestamp) Add(d native_time.Duration) Timestamp {
|
|
|
|
return t + Timestamp(d/MinimumTick)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Sub returns the Duration t - o.
|
|
|
|
func (t Timestamp) Sub(o Timestamp) native_time.Duration {
|
|
|
|
return native_time.Duration(t-o) * MinimumTick
|
|
|
|
}
|
|
|
|
|
|
|
|
// Time returns the time.Time representation of t.
|
|
|
|
func (t Timestamp) Time() native_time.Time {
|
|
|
|
return native_time.Unix(int64(t)/second, (int64(t) % second))
|
|
|
|
}
|
|
|
|
|
|
|
|
// Unix returns t as a Unix time, the number of seconds elapsed
|
|
|
|
// since January 1, 1970 UTC.
|
|
|
|
func (t Timestamp) Unix() int64 {
|
|
|
|
return int64(t) / second
|
|
|
|
}
|
|
|
|
|
|
|
|
// String returns a string representation of the timestamp.
|
|
|
|
func (t Timestamp) String() string {
|
|
|
|
return fmt.Sprint(int64(t))
|
|
|
|
}
|
|
|
|
|
|
|
|
// Now returns the current time as a Timestamp.
|
|
|
|
func Now() Timestamp {
|
|
|
|
return TimestampFromTime(native_time.Now())
|
|
|
|
}
|
|
|
|
|
|
|
|
// TimestampFromTime returns the Timestamp equivalent to the time.Time t.
|
|
|
|
func TimestampFromTime(t native_time.Time) Timestamp {
|
|
|
|
return TimestampFromUnix(t.Unix())
|
|
|
|
}
|
|
|
|
|
|
|
|
// TimestampFromUnix returns the Timestamp equivalent to the Unix timestamp t.
|
|
|
|
func TimestampFromUnix(t int64) Timestamp {
|
|
|
|
return Timestamp(t * second)
|
|
|
|
}
|